Intra-Atrial Conduction Delay Revealed by Multisite Incremental Atrial Pacing is an Independent Marker of Remodeling in Human Atrial Fibrillation

OBJECTIVES This study sought to characterize direction-dependent and coupling interval – dependent changes in left atrial conduction and electrogram morphology in uniformly classi ﬁ ed patients with paroxysmal atrial ﬁ brillation (AF) and normal bipolar voltage mapping. BACKGROUND Although AF classi ﬁ cations are based on arrhythmia duration, the clinical course, and treatment response vary between patients within these groups. Electrophysiological mechanisms responsible for this variability are incompletely described. METHODS Intracardiac contact mapping during incremental atrial pacing was used to characterize atrial conduction, activation dispersion, and electrogram morphology in 15 consecutive paroxysmal AF patients undergoing ﬁ rst-time pulmonary vein isolation. Outcome measures were vulnerability to AF induction at electrophysiology study and 2-year follow-up for arrhythmia recurrence. RESULTS Conduction delay showed a bimodal distribution, occurring at either long (high right atrium pacing: 326 (cid:1) 13 ms; coronary sinus pacing: 319 (cid:1) 16 ms) or short (high right atrium pacing: 275 (cid:1) 11 ms; coronary sinus pacing: 271 (cid:1) 11 ms) extrastimulus coupling intervals. Arrhythmia recurrence was found only in patients with conduction delay at long extrastimulus coupling intervals, and patients with inducible AF were characterized by increased activation dispersion (activation dispersion time: 168 (cid:1) 29 ms vs. 136 (cid:1) 11 ms). Electrogram voltage and duration varied throughout the left atrium, between patients, and with

A trial fibrillation (AF) classifications that are used to recommend treatment decisions are based on AF episode duration (1); however, a number of observations suggest that within classification categories, AF is a structurally and electrically diverse arrhythmia. First, ablation shows variable success between apparently similar patients when controlling for comorbidities and left atrial (LA) dimensions. Second, some patients require multiple ablation procedures to achieve freedom from AF despite durable pulmonary vein isolation, whereas others require only a single procedure. Finally, the natural history of AF varies, with differing rates of progression to persistent AF seen. In line with these observations, reduced conduction velocity and shortened refractoriness have been associated with increasing severity, duration or recurrence of AF in some (2), but not all (3,4), studies.
Recently, structural (5) and voltage-defined (6) targets for intervention beyond pulmonary vein isolation have been proposed. Although low voltage is correlated with the presence of magnetic resonance imaging indices of atrial fibrosis (7), preservation of normal bipolar voltage does not imply absence of fibrosis (8). Therefore, bipolar voltage measured only during sinus rhythm or fixed coupling interval pacing may fail to detect atrial structural change. Furthermore, whether LA conduction abnormalities or morphological electrogram changes occur in apparently healthy atria (with normal bipolar voltage) is unknown.
We hypothesized that direction-and coupling interval-dependent changes in electrogram morphology and timing in patients with apparently healthy atria may differentiate between truly normal atria and those with underlying substrate change. In this study, incremental atrial pacing was used to measure electrogram voltage (EV), duration, conduction, and activation dispersion in a group of uniformly classified patients with paroxysmal AF (PAF) and otherwise normal atria.

PATIENT SELECTION AND CLINICAL PROCEDURES.
Ethical approval was granted by the National Research Ethics Service (10/H0802/77), and all participants gave written informed consent for study inclusion. The research conformed to the principles described in the Declaration of Helsinki. Patients with ischemic heart disease, cardiac surgery, or structural heart disease were excluded. Antiarrhythmic drugs, including calcium channel blockers, were stopped at least 5 half-lives before ablation.
Amiodarone was stopped at least 6 weeks before ablation. All clinical procedures were performed under general anesthesia.
Following femoral access and trans-septal puncture, 2   interval illustrating measurement of S1S2 delay and S1S2 block times. Relationship between electrogram voltage (C) or electrogram duration (D) and extrastimulus coupling interval with study parameters marked. DED ¼ rate dependence of electrogram duration; ED ¼ electrogram duration; EV ¼ electrogram voltage; DEV ¼ rate dependence of electrogram voltage; S1S2 block ¼ shortest S1S2 conducted to the left atrial recording site; S1S2 delay ¼ shortest S1S2 interval that conducted without decrement to the left atrium.

DED
Maximal increase in ED at short extrastimulus coupling interval compared to baseline ED ms Conduction S1S2 delay The shortest S1S2 coupling interval conducting to the left atrium with S1S2 ¼ A1A2 ms S1S2 block The shortest S1S2 coupling interval that conducts from pacing site to left atrium ms ADT Longest LA activation time (measured at shortest S1S2 coupling interval) minus shortest LA activation time (measured at longest S1S2 coupling interval) ms Refractoriness ERP The longest S1S2 coupling interval at which S2 failed to produce local capture ms Williams et al.   Figure 1C). The rate dependence of electrogram duration was quantified from S1S2-A1A2 plots for each recording site.
For each S2 cycle length, the 5th and 95th centile of A2 electrogram components were taken as the beginning and end of local activation, respectively.
Change in electrogram duration was quantified as the longest duration between these curves minus the duration at baseline (DED, ms), which represents the maximal change in electrogram duration with coupling interval ( Figure 1D). Electrogram fractionation was quantified by counting the number of peaks/troughs greater than the noise threshold for each filtered electrogram. curves were plotted. These curves were characterized by: 1) an A1A2 equal to S1S2 at long pacing cycle lengths; 2) a minimum achievable A1A2 at short S1S2; and 3) a curved transition period between these regions. A hyperbola with asymptotes at y ¼ c and y ¼ x was fitted to these data. S1S2 block was defined as the shortest S1S2 conducted to the LA recording site. The shortest S1S2 interval that conducted without decrement to the left atrium (S1S2 delay ) was determined as the transition point of the curve (where A1A2 becomes greater than S1S2). S1S2 delay represents the coupling interval at which initial delay of the A2 electrogram occurs ( Figure 1B). A t r i a l r e f r a c t o r i n e s s . Atrial ERP was measured at 2 sites (HRA and mid-CS) at a single basic cycle length (470 ms). ERP was defined as the longest S1S2 extrastimulus coupling interval that failed to produce local atrial capture.
AF VULNERABILITY AND FOLLOW-UP. AF vulnerability was assessed during pacing from HRA and CS.
Sustained AF was defined as AF triggered by the pacing protocol and lasting for more than 30 s (9).
Where AF resolved to an organized tachycardia,      (Figure 6). Within a single case, measured S1S2 delay was uniform across the left atrium and was independent of activation direction (S1S2 delay during CS pacing vs. S1S2 delay during HRA pacing; R 2 ¼ 0.6674, p < 0.0001).
ACTIVATION DISPERSION. Conduction delay close to the pacing site was correlated with S1S2 delay (R 2 ¼ 0.83, p < 0.0001), but insufficient to explain the overall conduction delay seen in the left atrium  Figure 7B). Prolonged S1S2 delay was correlated with increased activation dispersion (R 2 ¼ 0.42 for correlation between ADT >60 mm and S1S2 delay ,   indicating that S1S2 delay measurement is feasible without AF induction.  Kaplan-Meier survival curves representing time to arrhythmia recurrence for patients with low and high S1S2 delay (defined from the bimodal distributions illustrated in Figure 6) are shown in Figure 8.
Although all arrhythmia recurrences were seen in patients with conduction delay occurring at long extrastimulus coupling intervals (long S1S2 delay ), the difference between the curves was nonsignificant by the log-rank test. In clinical studies of atrial refractoriness, ERP increases with both age and hypertension in the absence of AF (9,12). In the presence of AF, ERP prolongs compared with controls (13), but as AF progresses from paroxysmal to persistent, measured ERP universally shortens (14). Hence, although shortening of refractoriness is associated with more severe forms of AF, prolongation of refractoriness may prevent AF in at-risk disease states. The data presented here show significant variation in ERP between cases. In agreement with these findings, previous work has shown that shortening of refractoriness, but not dispersion of refractoriness, is greater in chronic AF compared with PAF (15). Our study extends these findings to show that even within this group of uniformly classified PAF patients, significant intercase differences in refractoriness are identifiable.

DISCUSSION
A spectrum of AF-related electrical remodeling may therefore be present in these patients.
ELECTRICAL CONDUCTION. Previous in vitro work shows that discontinuous conduction, at a cellular Values are mean AE SD unless otherwise specified.
AF ¼ atrial fibrillation; CS ¼ coronary sinus; HRA ¼ high right atrium; other abbreviations as in Table 1.

FIGURE 8 Kaplan-Meier Survival Analysis
Survival curves are shown with patients dichotomized into groups based on the S1S2 delay populations identified in Figure 6 under HRA pacing (A) and CS pacing (B) conditions.
Abbreviations as in Figures 1 and 2.
Williams et al.